songsenand
/
cgal
mirror of https://github.com/CGAL/cgal
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
55,238 Commits 3 Branches 156 Tags 959 MiB
Commit Graph

1169 Commits

Author SHA1 Message Date
Marc Glisse 3b3993aa19 Merge branch 'Kernel_d-rewrite-glisse-old' into Kernel_d-rewrite-glisse 
Conflicts:
	Installation/changes.html
 2014-06-23 17:13:29 +02:00
Marc Glisse dd4f968c48 Reorder some macro tests, no functional change. 2014-06-06 12:32:11 +02:00
Marc Glisse 37cab384f8 Opacify FPU support for XL C++. It accepts gcc's asm syntax with slight variations. 2014-05-30 10:34:42 +02:00
Marc Glisse ccca2e6e24 IBM XL C++ support for counting leading/trailing zeros. 
Patch by George Zagaris.
 2014-05-29 18:11:31 +02:00
Marc Glisse 0f31b49301 Make Gmpq interoperable with long long (required by eigen...). 2014-05-09 13:06:22 +02:00
Marc Glisse 897f585f12 Merge branch 'Kernel_d-rewrite-glisse-old' into Kernel_d-rewrite-glisse 
Need the Interval_nt/long interoperability for Eigen.
 2014-05-07 18:41:17 +02:00
Marc Glisse 89dd491a17 Merge remote-tracking branch 'origin/Number_types-Interval_with_long-glisse' 
Remove explicit interoperability between Interval_nt and int, so
Interval_nt becomes interoperable with every builtin type (in particular
long) via double.

Approved by the release manager
 2014-05-07 13:19:24 +02:00
Marc Glisse e5113a2c37 Don't use Boost.Operators for +-*/ of Gmpq. 
It shouldn't change the performance significantly (the time is spent in
malloc/free and the mpq_* calls), but at least I can follow the
(smaller) generated code.
 2014-05-05 20:06:22 +02:00
Marc Glisse b8bab2041a Remove the mixed Interval_nt x int operations. They are useless and 
cause ambiguity with the double operations.
 2014-05-05 18:19:08 +02:00
Laurent Rineau a8ff620f3f Fix the remaining warnings on my platform 
That was mostly "unused locally defined typedefs".
 2014-04-01 14:50:51 +02:00
Laurent Rineau 136b536b1a Fix more warnings 2014-03-27 11:34:40 +01:00
Laurent Rineau 13dd6aed26 Fix warnings 2014-03-26 14:29:36 +01:00
Marc Glisse 50aba6f3dc Microsoft needs to do something about their warnings... 2014-03-15 10:00:57 +01:00
Marc Glisse e25a284796 VC and min/max :-( 2014-03-15 09:41:47 +01:00
Marc Glisse 5c7ca614bf On win64, eigen wants to convert __int64 to whatever number type we are using. 2014-03-14 15:02:17 +01:00
Marc Glisse 39f1c4f589 At least on windows with the latest eigen, dummy_precision seems 
necessary.
 2014-03-14 14:43:32 +01:00
Marc Glisse 1241b2f14f Use boost::int128_type instead of __int128. This makes __extension__ 
useless, hence removed.
 2014-03-13 17:06:43 +01:00
Marc Glisse b257e85427 Rework test for __int128 availability. 2014-03-13 16:50:22 +01:00
Marc Glisse 6571a80fdd fix pedantic warnings. 2014-03-13 13:21:21 +01:00
Marc Glisse 3168c851a1 Merge branch 'Kernel_d-rewrite-glisse-old' into Kernel_d-rewrite-glisse 
Conflicts:
	Kernel_d/doc_tex/Kernel_d/kernel_representation_d.tex
	Kernel_d/doc_tex/Kernel_d_ref/Homogeneous_d.tex
	Kernel_d/doc_tex/Kernel_d_ref/Kernel.tex
	Kernel_d/doc_tex/Kernel_d_ref/main.tex

Resolved by dropping those files.
 2014-03-12 12:38:08 +01:00
Laurent Rineau 367314623d Add a script to detect wrong permissions, and fix permissions 
The script uses zsh, and lists regular files that have the
'executable' permission.

At the same time, I fix the permissions of several files.
 2014-02-06 18:49:09 +01:00
Marc Glisse 91090ec2b7 Merge remote-tracking branch 'origin/Number_types-powerpc-glisse' 
Tested in CGAL-4.4-Ic-57 (and CGAL-4.4-Ic-51 but only for ppc64_linux_gcc47).

Approved by the Release Manager.
 2013-12-21 11:09:18 +01:00
Laurent Rineau 7b5a8a388d merge branch origin/Number_types-new_delete_mismatch-GF 
Remove memory leaks detected by AddressSanitizer.
Tested in CGAL-4.4-Ic-54 and CGAL-4.4-Ic-55.
Approved by the release manager.
 2013-12-11 13:42:34 +01:00
Marc Glisse 8f2b5970e7 Fix FPU rounding mode on powerpc. 
The code was incorrect, we should have used _FPU_GETCW(CW)&3 to keep
only the rounding mode bits. However, glibc comes with an optimized
inline version of fegetround and fesetround so there is no point using
anything else.
 2013-12-07 01:41:02 +01:00
Sébastien Loriot 0fa2cacb60 remove doc_tex documentation (including developer manual) and old manual tools 2013-12-03 17:27:02 +01:00
Marc Glisse 7f1adbc485 Reformulate one comment about IA_opacify for vectors. 2013-11-27 17:57:43 +01:00
Marc Glisse 8ddf8255d5 Merge branch 'Kernel_d-rewrite-glisse-old' into Kernel_d-rewrite-glisse 
Hopefully the doc will be easier to build this way.
 2013-11-23 12:34:18 +01:00
Marc Glisse a1c21ae45f Improve asm for -m32 and with Intel's compiler. 2013-11-19 00:46:00 +01:00
Andreas Fabri 18b04bfb22 delete [] instead of delete after a new array 2013-11-18 17:24:16 +01:00
Marc Glisse fffddfb806 Finish the removal of the template :-( 2013-11-17 22:18:39 +01:00
Marc Glisse d128a72551 IA_opacify is not generic, don't make it a template. 
The code has become very specific to double, and we weren't using it for
vectors yet.
 2013-11-17 12:54:34 +01:00
Marc Glisse 726a7daf63 Work around gcc bug. Improve support for non-x86. 2013-11-17 11:24:15 +01:00
Marc Glisse 325a687888 Fix llvm ICE on inline asm. Also improve asm for gcc. 
It is strange, my experiments today don't give at all the same results I
remember from this summer...
 2013-11-16 22:46:33 +01:00
Sébastien Loriot da1791759c Merge branch 'Number_types-mpzf-glisse' 
Introduces a faster alternative to MP_Float and Gmpzf
    (see the documentation to see on which platforms it is available).
    When activated, it is used as default exact ring number type for exact
    evaluation of predicates (that only require ring operations).
    It is possible to switch back to the previous behavior by defining
    CGAL_DO_NOT_USE_MPZF

    Successfully tested in CGAL-4.4-Ic-29
    Approved by the Release Manager
 2013-11-15 16:38:16 +01:00
Sébastien Loriot d99344bf98 add a way to easily switch Mpzf off 2013-11-14 12:55:39 +01:00
Marc Glisse 06e0774812 Cleanup assertions. 2013-11-14 10:38:34 +01:00
Laurent Rineau ae4ff9bc05 Minor rephrasing about the exactness of Mpfz(double) 2013-11-13 11:15:27 +01:00
Sébastien Loriot d466a4ce22 add Mpzf in the user manual 2013-11-12 19:23:16 +01:00
Sébastien Loriot 9168030414 explicit that constructors are exact 2013-11-12 18:03:52 +01:00
Sébastien Loriot e92b19a8e4 explicit when the macro CGAL_HAS+MPZF is defined 2013-11-12 17:48:27 +01:00
Sébastien Loriot 79e4030f09 rename mpzf to Mpzf 2013-11-12 17:40:46 +01:00
Luis Peñaranda ad5144404e typo, is->are 2013-11-06 11:51:13 -02:00
Marc Glisse d940e39efc Hopefully make MSVC quiet. 2013-11-06 14:31:29 +01:00
Luis Peñaranda 78de4bb35c updated Gmpfr and Gmpfi documentation 
Gmpfi:
- it was added to the implementation section of the reference manual
  that the precision is decoupled from the mpfr precision.

Gmpfr:
- it said that mpfr handles four rounding modes; this is wrong, since
  mpfr 3 handles five. Now, it reads that Gmpfr supports these four
  rounding modes.
- a spurious star was removed from the reference manual.

NT user manual:
- it said that the precision of Gmpfi was inherited from Gmpfr; this was
  changed in the past and the sentence was corrected.
- a missing comma was added in the last section.
 2013-11-06 11:15:32 -02:00
Luis Peñaranda 4d7c940c2f use CGAL/config.h instead to CGAL/basic.h 
The only use of this include here is to detect the definition
CGAL_HAS_THREADS.
 2013-11-05 20:29:15 -02:00
Luis Peñaranda f19d779046 adjusting to coding conventions in Gmp* tests 
All the preprocessor definitions and template parameters starting with _
and a capital letter were removed.
 2013-11-05 17:41:43 -02:00
Luis Peñaranda cf0e0a0445 bug in f2f36a1db3 
I forgot to copy the pointers back to the Gmpfi.
 2013-11-05 17:23:05 -02:00
Luis Peñaranda f2f36a1db3 fix memory leak in some Gmpfi operators 
The leak was caused by the fact that some operators created an mpfi_t to
store the result and returned a Gmpfi constructed from that mpfi_t. The
problem was that it was impossible to free the memory used by that
number. The fix consisted in storing the result directly in a Gmpfi,
instead of a mpfi_t (this also results in a cleaner code).
 2013-11-04 17:39:12 -02:00
Marc Glisse 4e58c5ebd6 Fix bug in comparison with 0. 2013-11-02 11:35:06 +01:00
Marc Glisse b53de35ff6 Work around Boost changing the name of their macros every other release. 
Work around Microsoft defining min/max as macros.
 2013-11-01 14:46:20 +01:00